GdmRIII, a TetR Family Transcriptional Regulator, Controls Geldanamycin and Elaiophylin Biosynthesis in Streptomyces autolyticus CGMCC0516

Geldanamycin and elaiophylin are co-produced in several Streptomyces strains. However, the regulation of their biosynthesis is not fully understood yet. Herein the function of a TetR family regulator GdmRIII, which is located in the biosynthetic gene cluster of geldanamycin, was studied to understand the regulatory mechanism of geldanamycin biosynthesis in Streptomyces autolyticus CGMCC0516. The production of geldanamycin decreased substantially in a ΔgdmRIII mutant and the yield of three compounds which were thought to be geldanamycin congeners greatly increased. Surprisingly, the structural elucidation of these compounds showed that they were elaiophylin and its analogues, which implied that GdmRIII not only played a positive regulatory role in the biosynthesis of geldanamycin, but also played a negative role in elaiophylin biosynthesis. GdmRIII affected the expression of multiple genes in both gene clusters, and directly regulated the expression of gdmM, gdmN, and elaF by binding to the promoter regions of these three genes. A conserved non-palindromic sequence was found among the binding sites of elaF. Our findings suggested that the biosynthetic pathways of geldanamycin and elaiophylin were connected through GdmRIII, which might provide a way for Streptomyces to coordinate the biosynthesis of these compounds for better adapting to environment changes.

conserved helix-turn-helix DNA-binding domain and a larger C-terminal ligand-binding domain 11 . Many of these regulators bind to palindromic sequences upstream of the genes they regulate [12][13][14] , while others bind to non-palindromic sequences 15 . Most of the TetR family regulators are identified as homodimeric negative regulators, and only a small number act as activators [16][17][18][19][20] or repressors/activators 21,22 . Although there are more than 200,000 sequences of probable TetR family regulators in the public databases and the structure for close to 200 have been resolved, the vast majority of these regulators have not been characterized. The soil-dwelling bacteria, such as Actinobacteria, encode the highest numbers of TetR family regulators. Whole genome sequencing projects reveal that some Streptomyces species, including Streptomyces coelicolor and Streptomyces avermitilis, contain over 100 genes for TetR family regulators, which presumably reflects the complex morphological differentiation and secondary metabolism in these species 10 .
Benzoquinone ansamycin antibiotic geldanamycin was first found to be produced by Streptomyces hygroscopicus in 1970 23 , and attracted wide attention due to its antitumor activity 24,25 . Geldanamycin is usually co-produced with another antibiotic elaiophylin 26 , which represents a class of C 2 -symmetric 16-membered macrodiolide antibiotics with various biological activities [27][28][29][30][31] . The fact that these two compounds are co-produced is intriguing, since they are synthesized by disparate pathways. The biosynthetic gene clusters for both compounds have been characterized [32][33][34] . The geldanamycin biosynthetic gene cluster contains two families of regulators, the LuxR-family regulators GdmRI and GdmRII and the TetR-family regulator GdmRIII. Meanwhile, the elaiophylin biosynthetic gene cluster contains a postulated LuxR-family regulator and two putative two-component regulators. Whether there is a connection between the geldanamycin and elaiophylin biosynthetic pathways is currently unknown.
In our previous work, a geldanamycin-producing strain Streptomyces autolyticus CGMCC0516 was found to be able to produce a trace of autolytimycin, whose structure is closely related to that of geldanamycin 35 . Autolytimycin has been shown to bind the heat shock protein 90 (Hsp90) with enhanced binding activity than 17-allylamino-17-demethoxygeldanamycin (17-AAG), a geldanamycin derivative currently under evaluation for treatment of cancer 36 . Since autolytimycin is a by-product of the geldanamycin biosynthetic pathway, further understanding the regulation of this pathway would bring to light new approaches for increasing the production of autolytimycin through genetic engineering. In the present work, the function of GdmRIII in S. autolyticus CGMCC0516 was investigated. A gdmRIII knockout mutant was constructed and its fermentation products were analyzed; the expression of genes in the geldanamycin and elaiophylin biosynthetic gene clusters was examined by using reverse transcription PCR (RT-PCR) and quantitative real-time PCR; GdmRIII was overexpressed in Escherichia coli and its binding to the potential targets was analyzed by using electrophoretic mobility shift assays (EMSAs) and DNase I footprinting assays. We found that GdmRIII played a positive regulatory role in the biosynthesis of geldanamycin, while playing a negative role in elaiophylin biosynthesis. Thus, it acted as a pleiotropic regulator in the biosynthesis of secondary metabolites in S. autolyticus CGMCC0516.

Results
GdmRIII plays a positive regulatory role in the biosynthesis of geldanamycin. To explore the regulatory role of GdmRIII in the biosynthesis of geldanamycin, a gdmRIII gene knockout mutant was constructed by using the PCR-targeted gene replacement method as described 37 (See Supplementary Fig. S1a). An Apra R and Thio S double-crossover ΔgdmRIII mutant was recovered, and verified by PCR (See Supplementary Fig. S1b) and subsequently by Southern blot (See Supplementary Fig. S1c,d). The mutant did not show any obvious morphological change or growth defect (See Supplementary Fig. S2). The fermentation products of the ΔgdmRIII mutant were extracted and analyzed by using HPLC. The results revealed that the production of geldanamycin substantially decreased in the mutant when compared to that in the wild-type strain (Fig. 1a). The yield of geldanamycin in the mutant was only around 19% of that in the wild-type strain on day 7 (Fig. 1b). Meanwhile, the yields of three compounds with retention time of 24.4 min (compound 1), 26.6 min (compound 2), and 28.4 min (compound 3) increased to about 3.2 fold, 2.4 fold, and 1.7 fold of that in the wild-type strain, respectively (Fig. 1a,b). To exclude the polar effect of the replacement cassette on the nearby genes in the cluster, the wild-type gdmRIII gene was cloned in pUWL201apr, in which the expression of gdmRIII was under the control of a constitutive promoter ermE*, and introduced into the ΔgdmRIII mutant. The complementation analysis showed that the production of geldanamycin recovered to 80% of the regular level after introduction of a wild-type gdmRIII gene into the mutant. At the same time, the yields of the other three compounds almost returned to the levels of that in the wild-type strain (Fig. 1a,b), indicating that the changes of the production of these compounds were due to the mutation in gdmRIII. Thus, GdmRIII played a positive regulatory role in the biosynthesis of geldanamycin, which was consistent with the published report that a homologue of GdmRIII, named Gel19, positively control the biosynthesis of geldanamycin in S. hygroscopicus JCM4427 38 .
The secondary metabolites whose production increased in the ΔgdmRIII mutant are elaiophylin and its analogues. Our previous experiments showed that the production of three compounds, 1, 2, and 3, increased in the ΔgdmRIII mutant. We were interested in whether these compounds were intermediates of the geldanamycin biosynthetic pathway. Fermentation with the ΔgdmRIII mutant was carried out for 7 days, and compound 1, 2, and 3 were purified and their chemical structures were determined. Surprisingly, the ESI-MS analyses showed that the molecular weight of compound 1, 2,  Tables S1-S3). The spectra were consistent with those in the literature 28,29 . Therefore, GdmRIII also regulated the elaiophylin biosynthetic pathway, in a negative way. This was unexpected, since there is no report yet showing that GdmRIII, a member of the TetR family regulators situated in the geldanamycin gene cluster, could control the biosynthesis of elaiophylin.
We then tried to identify the genes whose expression was under the control of GdmRIII in the geldanamycin biosynthetic gene cluster by comparing their transcriptional levels in the ΔgdmRIII mutant to that in the wild-type strain. The transcription of genes involved in the biosynthesis of geldanamycin, including those encoding polyketide synthases (PKSs), tailoring enzymes, and regulators (the genetic organization of the geldanamycin gene clusters is illustrated in Fig. 2), was examined by using RT-PCR. The results showed that the transcription of genes involved in multiple processes of geldanamycin biosynthesis was affected by the ΔgdmRIII mutation. Specifically, the transcriptional levels of PKS gene gdmAI, tailoring enzyme genes gdmF, gdmM, gdmN, gdmH, gdmK, and gdmP, and regulatory genes gdmRI and gdmRII in the ΔgdmRIII mutant decreased or were abolished when compared to that in the wild-type strain on day 5 (See Supplementary Fig. S25). Quantitative real-time PCR was then used to further confirm the changes of the transcriptional levels of these genes, and the results showed that the transcription of gdmAI, gdmF, gdmM, gdmN, gdmH, gdmK, gdmP, gdmRI, and gdmRII indeed decreased in the ΔgdmRIII mutant (Fig. 3). Therefore, GdmRIII controlled, directly or indirectly, the expression of multiple genes in the geldanamycin biosynthetic pathway (illustrated in Fig. 2).

GdmRIII negatively regulates the expression of multiple genes in the elaiophylin gene cluster.
To find out whether GdmRIII negatively regulates the expression of genes in the elaiophylin gene cluster, RT-PCR was first carried out to examine the transcriptional levels of six major genes in the cluster, elaE, elaF, elaG, elaI, and elaO (the genetic organization of the elaiophylin clusters is illustrated in Fig. 2). The results showed that the transcription of elaE, elaF, elaG, and elaO were enhanced in the ΔgdmRIII mutant on day 5 compared to that in the wild-type strain (See Supplementary Fig. S26). Therefore, the expression of these genes might be negatively controlled by GdmRIII (illustrated in Fig. 2). However, the transcription of elaI was a little weakened in the ΔgdmRIII mutant. Quantitative real-time PCR was then performed to further verify the effect of the ΔgdmRIII mutation on the expression of above genes. The results confirmed that the transcription of genes elaE, elaF, elaG, and elaO all increased in the mutant compared to that in the wild-type strain, while the transcription of elaI, which encodes for an exporter, decreased slightly (Fig. 4). The increase of expression of the major genes related to elaiophylin biosynthesis was correlated with the yield increase of elaiophylin and its analogues in the ΔgdmRIII mutant. Therefore, GdmRIII regulated two disparate biosynthetic pathways of secondary metabolites in S. autolyticus CGMCC0516, in an antagonistic mode.
GdmRIII binds to the promoter regions of genes gdmM, gdmN, and elaF in vitro. To examine whether GdmRIII binds to the promoter regions of genes whose expression was affected by GdmRIII to directly regulate their expression, GdmRIII protein was overexpressed in E. coli and purified. The binding of GdmRIII to the probable promoters of genes in the geldanamycin and elaiophylin gene clusters was then examined by using EMSAs. SDS-PAGE analysis of the purified GdmRIII showed that the molecular weight of His 6 GdmRIII was 27.9 kDa as anticipated (See Supplementary Fig. S27a). The purified GdmRIII was further verified by western blot (See Supplementary Fig. S27b).
The binding of GdmRIII to the probable promoter regions of genes in the geldanamycin and elaiophylin gene clusters was then analyzed. The upstream regions of genes gdmAI, gdmF, gdmM, gdmN, gdmH, gdmK, gdmP, gdmRI, and gdmRII in the geldanamycin cluster and elaE, elaF, elaG, elaI, and elaO in the elaiophylin cluster (illustrated in Fig. 2) were amplified from the genomic DNA of the wild-type strain. EMSAs were carried out, and the results showed that GdmRIII could bind to the promoter regions of gdmM, gdmN, and elaF in vitro (Fig. 5). Competitive EMSAs indicated that the binding could be easily competed by 50-fold of specific DNA, however not by 50-fold of nonspecific DNA. The binding of GdmRIII to the promoter regions of these genes further confirmed the direct regulatory role of GdmRIII on them. However, no binding of GdmRIII protein to the promoter regions of other genes was found, which might be due to the non-optimal conditions used in EMSAs, lack of necessary cellular factors, or the possibility that their expression was not under the direct control of GdmRIII.  GdmRIII binds to a conserved non-palindromic sequence. In order to precisely pinpoint the binding sites of GdmRIII, DNase I footprinting assays were performed to identify the promoter sequence of elaF protected by GdmRIII. The results showed that the regions protected by GdmRIII were −137 to −118 nt (5′-CACCATGATGGAGGACCACT-3′) and −102 to −83 nt (5′-AGGCCATCGAGGACTGGCTG-3′) relative to the translation start codon of elaF (Fig. 6). The changes did not look very significant, which might be caused by weak binding of GdmRIII to its targets under the conditions used for DNase I footprinting. A conserved sequence (5′-ATNGAGGAC-3′) was found among the binding sites. However, there was no apparent palindromic sequence in the binding sites, which is often recognized by TetR family regulators 8 . Nevertheless, our results demonstrated that GdmRIII could directly bind to the promoter regions of genes in the geldanamycin and elaiophylin gene clusters to regulate their expression, and provided another example of the diversity among operator sequences of the TetR family regulators, which are often difficult to accurately predict 8 .

Discussion
In the present work, the TetR family regulator GdmRIII was found to not only control the biosynthesis of geldanamycin but also regulate the biosynthesis of elaiophylin and its analogues, acting as a pleiotropic regulator in the biosynthesis of bioactive secondary metabolites in S. autolyticus CGMCC0516.
A homologous protein of GdmRIII (named Gel19) is reported as a positive regulator in the biosynthesis of geldanamycin in S. hygroscopicus JCM4427 38 , which was confirmed by our analyses. However, the ΔgdmRIII mutant of S. autolyticus still produced geldanamycin, albeit substantially less, unlike that of S. hygroscopicus JCM4427, where production of geldanamycin is totally abolished. Accordingly, the expression of gdmAI was detected, though greatly decreased, in the ΔgdmRIII mutant but not in the Δgel19 mutant. Furthermore, the expression of post-modification gene gdmN (gel8) and the LuxR regulator gene gdmRI (gel17) decreased in the ΔgdmRIII mutant, contrary to that reported in S. hygroscopicus JCM4427. The discrepancy might be due to the different background of strains used. The expression of other genes, such as gdmP (gel16) and gdmRII (gel14), either decreased or was abolished, which was consistent with that reported 38 . In addition, we found that the expression of some other biosynthetic genes, gdmF, gdmM, gdmH, and gdmK, were all affected by the ΔgdmRIII mutation. Their expression is not examined in the reported work 38 . The effect of the ΔgdmRIII mutation on these genes might be indirect, since the mutation caused decreases in the transcriptional levels of two LuxR-family regulator genes, gdmRI and gdmRII, which are known to positively control the expression of polyketide biosynthetic related genes 38,39 .
The intergenic regions between gdmRIII and adjacent genes are less than 200 bp, which is consistent with that in most of the TetR regulators. The DNase I footprinting assays showed that GdmRIII bound to two sites in the promoter region of elaF. No palindromic motifs were found in the binding sites, which is also observed with another TetR family regulator AtrA that regulates the biosynthesis of daptomycin in Streptomyces roseosporus 15 . This further illustrated the reported diversity of operator sequences of TetR family regulators 10 .
The elaiophylin gene cluster and its biosynthetic pathway are previously reported 34 , however little is known about the regulation of this pathway. In our work, GdmRIII was found also regulating the biosynthesis of elaiophylin. Our results showed that the transcriptional levels of related genes elaE (Type I PKS), elaF (Type II thioesterase), elaG (glycosytransferase), and elaO (NDP-hexose 2,3-dehydratase) increased in the ΔgdmRIII mutant when compared to that in the wild-type strain, suggesting GdmRIII negatively controlled their expression. The EMSA analyses indicated that GdmRIII bound to the promoter region of elaF, therefore GdmRIII might directly control the expression of elaF, whose function is proposed to unblock PKS modules and restore overall efficiency of the complex enzyme 40 . To the best of our knowledge, this is the first report that the biosynthesis of elaiophylin is under the control of GdmRIII, which resides within the gene cluster of geldanamycin biosynthesis.
Elaiophylin is usually co-produced with geldanamycin 26 , but whether there is an underlying relationship between the two pathways is unknown. The fact that GdmRIII simultaneously regulated these two pathways in S. autolyticus CGMCC0516 suggested that the two pathways were actually connected, at least partially, through GdmRIII. The control of GdmRIII over these two disparate pathways is interesting, since being a cluster situated regulator, GdmRIII was thought to be pathway-specific, but not pleiotropic. The cross-regulation of disparate biosynthetic pathways we found with GdmRIII, resonated with a few other reports 6,16,[41][42][43] , further revealed the complexity of regulation that governs the biosynthesis of secondary metabolites in Streptomyces. It has long been recognized that the production of multiple secondary metabolites is coordinated during the growth cycle of Streptomyces, perhaps to facilitate their ability to compete against other biological species 44 . While the biological significance of cross-regulation exerted by GdmRIII is unknown, it might be a way for cells to coordinate disparate biosynthetic pathways for better adapting to environmental changes.
GdmRIII controlling the biosynthetic pathways of geldanamycin and elaiophylin in an antagonistic mode was rather intriguing. A similar regulatory pattern is also observed with RrdA in S. coelicolor, in which an rrdA null mutant exhibits increased undecylprodigiosin production and decreased actinorhodin production 45 . However, different from that with GdmRIII, the gene encoding RrdA resides within neither of these two gene clusters, and its influence on the biosynthesis of actinorhodin is supposed to be exerted at a non-transcriptional level. Biosynthesis of secondary metabolites is highly regulated in Streptomyces, which is associated with specific stages of the growth cycle, and often occurs during nutrient limitation. The supply of metabolites utilized in common by different pathways has been postulated to alter the actions of pathways whose protein products compete for the same precursor 46 . Since malonyl-CoA and methylmalonyl-CoA are both used as the extension units for the biosynthesis of geldanamycin 47 and elaiophylin 48 , the competition between these two pathways for pools of precursor metabolites used in common might be the reason they were regulated in an antagonistic mode, which in turn might confer an adaptive advantage by selectively synthesizing the needed compound during nutrient limitation. It is worth noting that the presence or lack of GdmRIII did not completely shut down the production of elaiophylin or geldanamycin, which is consistent with the fact that these two compounds are usually co-produced. Other unknown cellular factors must have dictated the biosynthesis of geldanamycin and elaiophylin, while GdmRIII works to fine tune their relative amount to exert more elaborate control over the processes. Further studies on the distinct role of GdmRIII played in disparate biosynthetic pathways would help us to understand the complex control of secondary metabolism in Streptomyces.

Methods
Strains, plasmids, and growth conditions. Wild-type strain S. autolyticus CGMCC0516 has been described previously 35 . Escherichia coli BW25113/pIJ790 37 was used for PCR-targeting and λ-RED-mediated mutagenesis, ET12567/pUZ8002 49 for intergeneric conjugation, DH5α 50 for routine cloning, and BL21(DE3) 51 for protein expression. Cosmid pSA2065, which contains the wild-type gdmRIII gene, was used for constructing the gene knockout. pHY773, a derivative of pIJ773 37 in which the oriT fragment has been deleted, was used as the template for amplifying the replacement cassette. pUWL201apr, which contains the constitutive promoter ermE* Figure 6. Determination of the binding sites of of GdmRIII. (a) DNase I footprinting assay of the elaF promoter protected by GdmRIII. The middle electropherogram represents the sequencing reaction without GdmRIII protein and the lower electropherogram represents the reaction with GdmRIII protein added. The regions where the blue peaks are constantly lower than the red peaks are considered to be protected by GdmRIII; (b) the nucleotide sequence of the elaF promoter region and the GdmRIII binding sites. The numbers indicate the distance (nt) from the translational start point. The solid lines indicate the GdmRIII binding sites. (c) the conserved sequence among GdmRIII binding sites. upstream of the multiple cloning site, was used for constructing the complementation plasmid. pEASYT1 (Trans, Beijing) was used as the T-cloning vector, and pET29a (Novagen, Madison) for protein expression.
Gene disruption and complementation. DNA isolation and manipulation in E. coli and Streptomyces were carried out according to standard procedures 49,50 . A ΔgdmRIII mutant strain was constructed as described 37 . The gdmRIII gene disruption cassette was PCR amplified from pHY773 with primers 5′-TATTGGCTGACAGCCAGCCAACGCAGGAGTTACAGCATGATTCCGGGGATCCGTCGACC-3′ and 5′-GGGCCCGTCCCTCACCCATGAGTCACCTCTGAGTGCTCATGTAGGCTGGAGCTGCTTC-3′, and introduced into BW25113/pIJ790/pSA2065 by electroporation. The strain harboring the disrupted gdmRIII gene on pSA2065 was screened on LB plates supplemented with apramycin. The resultant cosmid, named pSRIII, was introduced into ET12567/pUZ8002 by electroporation, and then conjugation between ET12567/pUZ8002/ pSRIII and S. autolyticus CGMCC0516 was carried out as described 49 . The double-crossover mutants were identified by screening for Apra R and Thio S colonies. The mutants were verified by PCR amplification with primers 5′-GGAGCATCTCCCGGTGGACCG-3′ and 5′-CCGCTCTCCGGCGACACGCTCA-3′, and by Southern blot 50 using a DIG High Prime DNA Labeling and Detection Starter Kit I (Roche, Basel).
For complement at ion ana lysis, gdmRIII was amplif ie d f rom wi ld-typ e S. autoly ticus by using primers 5′ -CCCAAGCT TATGGTCCCCCGAAGCCCGTCGGTCAAT-3′ an d 5′-GGAATTCTCATCGACTCCCGTCGGGCGGC-3′. The amplicon was cloned into pUWL201apr and then introduced into the ΔgdmRIII mutant by conjugation to obtain the Amp R Thio R complementation strain HB-gdmRIII.
Fermentation and HPLC analysis of compounds. Streptomyces strains were grown on ISP4 plates at 28 °C for 72 hr, inoculated into TSB medium, and further incubated at 28 °C, 180 rpm for 36 hr to obtain seeds. The seeds were then inoculated at a ratio of 1:100 into 250 mL Fernbach flasks, each containing 50 mL of 61# medium. After incubating at 28 °C, 180 rpm for 168 hr, the liquid and mycelia were separated by centrifugation, and extracted with ethyl acetate and acetone, respectively. The extracts were combined, dried out with a rotary evaporation apparatus, and dissolved with methanol. The samples were analyzed by using an Agilent 1200 liquid chromatograph system with a Thermo C 18 column (5 μm, 250 × 4.6 mm), which was equilibrated with 70% solvent A (H 2 O) and 30% solvent B (methanol), developed with a linear gradient (0-25 min, from 30% B to 100% B), then kept at 100% B for 5 min at a flow rate of 1 mL/min, and UV detected at 254 nm. Three independent fermentations were carried out, and compounds isolated from each fermentation were analyzed with three technical replicates.
Reverse transcription PCR and real-time PCR analyses of gene expression. Total RNA was isolated from strains cultured for 120 hr in 61# medium with a RNAprep Pure Cell/Bacteria kit (TIANGEN, Beijing), and reverse transcription was carried out by using a PrimeScript RT kit with gDNA Eraser (Takara, DaLian) according to the manufacturer's instruction. PCR was performed with primers shown in Supplementary Table S4 at the following condition: 94 °C 30 sec, 62 °C 30 sec, and 72 °C 60 sec, for a total of 30 cycles. Samples of each WT and mutant gene pair at certain time point were run on the same gel. The images were captured by using Syngene GeneGenius Gel Imaging System with the same default settings, and the relative expression levels were evaluated by using Genesnap. RNA samples used in real-time PCR were isolated from fermentation cultures incubated for 120 hr. The amplification was carried out on a ABI StepOne Plus system using a SuperReal PreMix Plus kit (TIANGEN, Beijing), with primers listed in Supplementary Table S5 at the following condition: 95 °C 15 min, 1 cycle, followed by 40 cycles of 95 °C 10 sec, and 60 °C 32 sec. Relative standard curve method was used for analyzing the difference in transcription of genes, with 16S rRNA as the internal control. For each assay, RNAs were isolated from three independent fermentation cultures, and the analyses were conducted with three technical replicates.

Mapping of the binding sites of GdmRIII by DNase I footprinting. The promoter region
of elaF that contains the probable binding sites of GdmRIII was amplified from the chromosomal DNA of wild-type S. autolyticus with primers 5′-AGGAGCTGCCCGACTGGCGGTCCTTCTG-3′ and 5′-AGTCGCTCCAGCCAGTCCTCGATGGCC-3′, and then cloned into pEASYT1. The resultant plasmids were used as the templates for preparing fluorescent FAM labeled probes. The probes were amplified with Dpx DNA polymerase (TOLO Biotech, Shanghai), by using primers of M13F-47 (FAM) (5′-CGCCAGGGTTTTCCCAGTCACGAC -3′) and M13R-48 (5′-AGCGGATAACAATTTCACACAGGA-3′). The FAM-labeled probes were purified by using a Wizard ® SV Gel and PCR Clean-Up System (Promega, Madison) according to the manufacturer's instruction, and quantified with NanoDrop 2000C. DNase I footprinting assays were performed as described 52 , except that 400 ng of probes and 1.6 µM of GdmRIII proteins were used in the assays. Data availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.